Bankruptcy prediction is a finance problem that has been studied by researchers since the early nineteenth century. Bankruptcy prediction is an important problem in finance because predicting bankruptcy in advance allows stakeholders, such as lenders, stock and bond investors, and public authorities, to take early actions to limit the economic losses. In this paper, we present a summary of the bankruptcy prediction models, starting with the univariate statistical models to the latest cutting-edge models using machine learning and deep learning techniques. This study summarizes the size of training data, features, algorithms, and performance metrics used by machine learning and deep learning models. This study will help researchers, academia, and industry participants to understand the evolution of bankruptcy prediction models from statistical models to deep learning models. This study, for the first time, classifies the evolution of bankruptcy prediction models into phases for ease of understanding. We also present an in-depth analysis of the trends noticed in each of these phases.